Ex-OpenAI Policy Chief Brundage Rebuts Altman: OpenAI and Anthropic Barely Differ

Miles Brundage, former OpenAI head of policy and safety, posted a series of responses to Sam Altman's POLITICO interview, with a core conclusion: OpenAI and Anthropic do not have the major divergence in worldviews on AI regulation that Altman claimed — the two companies are essentially nearly identical, and both are willing to accept elevated catastrophic risk to keep racing.

Confirmed

• In the POLITICO interview, Altman said OpenAI and Anthropic have very different worldviews on AI regulation, citing Anthropic's support for a licensing regime for non-frontier models; Brundage said that when the reporter pressed for a specific example, Altman's cited example was completely inaccurate.
• Brundage argued that both companies' ideological positions are far from most of the world's, and essentially nearly identical.
• He criticized the stance of "accepting bad things happening so the technology can reach the world as fast as possible," noting that Anthropic itself knowingly continues to deploy Claude despite abuse (including use for ransomware and authoritarian propaganda in places like Iran, North Korea, and China).
• On regulatory thresholds, Brundage argued Anthropic's problem runs the opposite way: possibly unwilling to clash head-on with a16z, it endorsed a revenue threshold that "experts almost uniformly consider stupid"; he also commented on the interviewer's questioning style.

Why it matters

• This dispute unfolds as OpenAI faces scrutiny over recent events, and Brundage finds it especially glaring that "sloppiness/laissez-faire" is being packaged as a worldview virtue at this moment.
• As OpenAI's former internal policy lead, Brundage's public rebuttal directly challenges the two leading companies' narrative of differentiating themselves via "divergence," highlighting the shared tension between safety and the commercialization race at frontier labs.

2026-10-05 ~ 2026-10-05 · 7 related posts

• [source] Brundage pushes back on Altman: OpenAI and Anthropic have little real daylight — Miles_Brundage · 2026-10-05
• Brundage: Anthropic keeps deploying Claude despite ransomware and propaganda misuse — Miles_Brundage · 2026-10-05
• Ex-OpenAI exec Miles Brundage slams Claude's misuse by sanctioned states — Miles_Brundage · 2026-10-05
• [source] Brundage: OpenAI and Anthropic are basically identical, both racing with high catastrophe risk — Miles_Brundage · 2026-10-05
• [source] Brundage: Altman's Politico interview example was false, OAI and Anthropic nearly identical — Miles_Brundage · 2026-10-05
• Miles Brundage: Anthropic backs an AI revenue regulatory threshold experts agree is 'dumb' — Miles_Brundage · 2026-10-05

1 near-duplicate retellings: Miles_Brundage